Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Classification Algorithms”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 577 records · Page 32Linked to original sources
PubMed1991

Use of immunoglobulin heavy-chain and light-chain measurements in a multicenter trial to investigate monoclonal components: II. Classification by use of computer-based algorithms.

We describe a computer algorithm for classifying serum monoclonal proteins (MC) based on serum protein electrophoresis (SPE) and the automated measurement of kappa and lambda light chains and IgG, IgA, and IgM. We developed the algorithm by using a large database of unselected samples containing MC collected in a multicenter study. The performance of the algorithm was optimized by using iterative computational procedures and was tested on both the development database and on an independent set of MC-containing samples. With the development database, the algorithm correctly classified 50% and misassigned 2.5% of the MC. Where the MC were present in concentrations greater than 10 g/L, the rate of successful classification increased to 72% with 3% misclassification. When the algorithm was tested on a group of 101 MC-containing samples from an independent source, 67% were correctly classified and 8% misclassified, half of the latter being unusual IgD myelomas. We discuss the scope for the application of the algorithm in routine laboratory practice involving personal computer software.

Algorithms↗
PubMed2004

Clinical trial design for microarray predictive marker discovery and assessment.

Transcriptional profiling technologies that simultaneously measure the expression of thousands of mRNA species represent a powerful new clinical research tool. Similar to previous laboratory analytical methods including immunohistochemistry, PCR and in situ hybridization, this new technology may also find its niche in routine diagnostics. Outcome predictors discovered by these methods may be quite different from previous single-gene markers. These novel tests will probably combine the information embedded in the expression of multiple genes with mathematical prediction algorithms to formulate classification rules and predict outcome. The performance of machine learning-algorithm-based diagnostic tests may improve as they are trained on larger and larger sets of samples, and several generations of tests with improving accuracy may be introduced sequentially. Several gene-expression profiling-technology platforms are mature enough for clinical testing. The most important next step that is needed for further progress is the development and validation of multigene predictors in prospectively designed clinical trials to determine the true accuracy and clinical value of this new technology. This manuscript reviews methodological and statistical issues relevant to clinical trial design to discover and validate multigene predictors of response to therapy.

Clinical Trials as Topic↗
PubMed2003

Soft learning vector quantization.

Learning vector quantization (LVQ) is a popular class of adaptive nearest prototype classifiers for multiclass classification, but learning algorithms from this family have so far been proposed on heuristic grounds. Here, we take a more principled approach and derive two variants of LVQ using a gaussian mixture ansatz. We propose an objective function based on a likelihood ratio and derive a learning rule using gradient descent. The new approach provides a way to extend the algorithms of the LVQ family to different distance measure and allows for the design of "soft" LVQ algorithms. Benchmark results show that the new methods lead to better classification performance than LVQ 2.1. An additional benefit of the new method is that model assumptions are made explicit, so that the method can be adapted more easily to different kinds of problems.

Costs and Cost Analysis↗
PubMed1998

Automatic detection of stable and unstable chromosome aberrations visualized by three-color imaging after fluorescence in situ hybridization with a painting and a pancentromeric DNA probe.

Two-color fluorescence in situ hybridization (FISH) in combination with digital image analysis was used to develop an automatic system for the detection and classification of chromosome aberrations. Algorithms were developed for the automatic thresholding of the three digitized images: an FITC image representing specific painted chromosomes, a TRITC image representing the centromeres of all chromosomes, and a DAPI image representing all the counterstained chromosomes. A further algorithm was developed for the automatic classification of the different types of chromosome aberrations, such as translocations, dicentrics, and fragments. For this study, a dataset of 252 metaphases were digitized and analyzed automatically as well as manually. Of these metaphases, 81.3% could be correctly classified by the algorithm. The error rate was reduced to 9.3% by automatically excluding the detected clusters and artifacts. The average analysis time per metaphase was 34.5 s without any user intervention.

Adenocarcinoma↗
PubMed2005

Data classification with radial basis function networks based on a novel kernel density estimation algorithm.

This paper presents a novel learning algorithm for efficient construction of the radial basis function (RBF) networks that can deliver the same level of accuracy as the support vector machines (SVMs) in data classification applications. The proposed learning algorithm works by constructing one RBF subnetwork to approximate the probability density function of each class of objects in the training data set. With respect to algorithm design, the main distinction of the proposed learning algorithm is the novel kernel density estimation algorithm that features an average time complexity of O(n log n), where n is the number of samples in the training data set. One important advantage of the proposed learning algorithm, in comparison with the SVM, is that the proposed learning algorithm generally takes far less time to construct a data classifier with an optimized parameter setting. This feature is of significance for many contemporary applications, in particular, for those applications in which new objects are continuously added into an already large database. Another desirable feature of the proposed learning algorithm is that the RBF networks constructed are capable of carrying out data classification with more than two classes of objects in one single run. In other words, unlike with the SVM, there is no need to resort to mechanisms such as one-against-one or one-against-all for handling datasets with more than two classes of objects. The comparison with SVM is of particular interest, because it has been shown in a number of recent studies that SVM generally are able to deliver higher classification accuracy than the other existing data classification algorithms. As the proposed learning algorithm is instance-based, the data reduction issue is also addressed in this paper. One interesting observation in this regard is that, for all three data sets used in data reduction experiments, the number of training samples remaining after a naive data reduction mechanism is applied is quite close to the number of support vectors identified by the SVM software. This paper also compares the performance of the RBF networks constructed with the proposed learning algorithm and those constructed with a conventional cluster-based learning algorithm. The most interesting observation learned is that, with respect to data classification, the distributions of training samples near the boundaries between different classes of objects carry more crucial information than the distributions of samples in the inner parts of the clusters.

Algorithms↗
PubMed2006

Training reformulated radial basis function neural networks capable of identifying uncertainty in data classification.

This paper introduces a learning algorithm that can be used for training reformulated radial basis function neural networks (RBFNNs) capable of identifying uncertainty in data classification. This learning algorithm trains a special class of reformulated RBFNNs, known as cosine RBFNNs, by updating selected adjustable parameters to minimize the class-conditional variances at the outputs of their radial basis functions (RBFs). The experiments verify that quantum neural networks (QNNs) and cosine RBFNNs trained by the proposed learning algorithm are capable of identifying uncertainty in data classification, a property that is not shared by cosine RBFNNs trained by the original learning algorithm and conventional feed-forward neural networks (FFNNs). Finally, this study leads to a simple classification strategy that can be used to improve the classification accuracy of QNNs and cosine RBFNNs by rejecting ambiguous feature vectors based on their responses.

Algorithms↗
PubMed2006

Bayesian Gaussian process classification with the EM-EP algorithm.

Gaussian process classifiers (GPCs) are Bayesian probabilistic kernel classifiers. In GPCs, the probability of belonging to a certain class at an input location is monotonically related to the value of some latent function at that location. Starting from a Gaussian process prior over this latent function, data are used to infer both the posterior over the latent function and the values of hyperparameters to determine various aspects of the function. Recently, the expectation propagation (EP) approach has been proposed to infer the posterior over the latent function. Based on this work, we present an approximate EM algorithm, the EM-EP algorithm, to learn both the latent function and the hyperparameters. This algorithm is found to converge in practice and provides an efficient Bayesian framework for learning hyperparameters of the kernel. A multiclass extension of the EM-EP algorithm for GPCs is also derived. In the experimental results, the EM-EP algorithms are as good or better than other methods for GPCs or Support Vector Machines (SVMs) with cross-validation.

Algorithms↗
PubMed2004

Computer-aided diagnosis of mammographic microcalcification clusters.

Computer-aided diagnosis techniques in medical imaging are developed for the automated differentiation between benign and malignant lesions and go beyond computer-aided detection by providing cancer likelihood for a detected lesion given image and/or patient characteristics. The goal of this study was the development and evaluation of a computer-aided detection and diagnosis algorithm for mammographic calcification clusters. The emphasis was on the diagnostic component, although the algorithm included automated detection, segmentation, and classification steps based on wavelet filters and artificial neural networks. Classification features were selected primarily from descriptors of the morphology of the individual calcifications and the distribution of the cluster. Thirteen such descriptors were selected and, combined with patient's age, were given as inputs to the network. The features were ranked and evaluated for the classification of 100 high-resolution, digitized mammograms containing biopsy-proven, benign and malignant calcification clusters. The classification performance of the algorithm reached a 100% sensitivity for a specificity of 85% (receiver operating characteristic area index Az = 0.98 +/- 0.01). Tests of the algorithm under various conditions showed that the selected features were robust morphological and distributional descriptors, relatively insensitive to segmentation and detection errors such as false positive signals. The algorithm could exceed the performance of a similar visual analysis system that was used as basis for development and, combined with a simple image standardization process, could be applied to images from different imaging systems and film digitizers with similar sensitivity and specificity rates.

Algorithms↗
PubMed2006

Classification of breast cancer using genetic algorithms and tissue microarrays.

PURPOSE: A multitude of breast cancer mRNA profiling studies has stratified breast cancer and defined gene sets that correlate with outcome. However, the number of genes used to predict patient outcome or define tumor subtypes by RNA expression studies is variable, nonoverlapping, and generally requires specialized technologies that are beyond those used in the routine pathology laboratory. It would be ideal if the familiarity and streamlined nature of immunohistochemistry could be combined with the rigorously quantitative and highly specific properties of nucleic acid-based analysis to predict patient outcome. EXPERIMENTAL DESIGN: We have used AQUA-based objective quantitative analysis of tissue microarrays toward the goal of discovery of a minimal number of markers with maximal prognostic or predictive value that can be applied to the conventional formalin-fixed, paraffin-embedded tissue section. RESULTS: The minimal discovered multiplexed set of tissue biomarkers was GATA3, NAT1, and estrogen receptor. Genetic algorithms were then applied after division of our cohort into a training set of 223 breast cancer patients to discover a prospectively applicable solution that can define a subset of patients with 5-year survival of 96%. This algorithm was then validated on an internal validation set (n=223, 5-year survival=95.8%) and further validated on an independent cohort from Sweden, which showed 5-year survival of 92.7% (n=149). CONCLUSIONS: With further validation, this test has both the familiarity and specificity for widespread use in management of breast cancer. More generally, this work illustrates the potential for multiplexed biomarker discovery on the tissue microarray platform.

Algorithms↗
PubMed2006

In silico prediction of yeast deletion phenotypes.

Analysis of gene deletions is a fundamental approach for investigating gene function. We evaluated an algorithm that uses classification techniques to predict the phenotypic effects of gene deletions in yeast. We used a modified simulated annealing algorithm for feature selection and weighting. The selected features with high weights were phylogenetic conservation scores for bacteria, fungi (excluding Ascomycota), Ascomycota (excluding Saccharomyces cerevisiae), plants, and mammals, degree of paralogy, and number of protein-protein interactions. Classification was performed by weighted k-nearest neighbor and with support vector machine algorithms. To demonstrate how this approach might complement existing experimental procedures, we applied our algorithm to predict essential genes and genes causing morphological alterations in yeast.

Algorithms↗
PubMed2004

Multivariate selection of genetic markers in diagnostic classification.

Analysis of gene expression data obtained from microarrays presents a new set of challenges to machine learning modeling. In this domain, in which the number of variables far exceeds the number of cases, identifying relevant genes or groups of genes that are good markers for a particular classification is as important as achieving good classification performance. Although several machine learning algorithms have been proposed to address the latter, identification of gene markers has not been systematically pursued. In this article, we investigate several algorithms for selecting gene markers for classification. We test these algorithms using logistic regression, as this is a simple and efficient supervised learning algorithm. We demonstrate, using 10 different data sets, that a conditionally univariate algorithm constitutes a viable choice if a researcher is interested in quickly determining a set of gene expression levels that can serve as markers for disease. We show that the classification performance of logistic regression is not very different from that of more sophisticated algorithms that have been applied in previous studies, and that the gene selection in the logistic regression algorithm is reasonable in both cases. Furthermore, the algorithm is simple, its theoretical basis is well established, and our user-friendly implementation is now freely available on the internet, serving as a benchmarking tool for the development of new algorithms.

Algorithms↗
PubMed1996

Lossless medical image compression by multilevel decomposition.

Lossless image coding is important for medical image compression because any information loss or error caused by the image compression process could affect clinical diagnostic decisions. This paper proposes a lossless compression algorithm for application to medical images that have high spatial correlation. The proposed image compression algorithm uses a multi-level decomposition scheme in conjunction with prediction and classification. In this algorithm, an image is divided into four subimages by subsampling. One subimage is used as a reference to predict the other three subimages. The prediction errors of the three subimages are classified into two or three groups by the characteristics of the reference subimage, and the classified prediction errors are encoded by entropy coding with corresponding code words. These subsampling and classified entropy coding procedures are repeated on the reference subimage in each level, and the reference subimage in the last repetition is encoded by conventional differential pulse code modulation and entropy coding. To verify this proposed algorithm, it was applied to several chest radiographs and computed tomography and magnetic resonance images, and the results were compared with those from well-known lossless compression algorithms.

Algorithms↗
PubMed2003

Quantitative assessment of carotid plaque composition using multicontrast MRI and registered histology.

MRI is emerging as a promising modality for monitoring carotid atherosclerosis. Multiple MR contrast weightings are required for identification of plaque constituents. In this study, eight MR contrast weightings with proven potential for plaque characterization were used to image carotid endarterectomy specimens. A classification technique was developed to create a tissue-specific map by incorporating information from all MR contrast weightings. The classifier was validated by comparison with micro-CT (calcification only) and with matched histological slices registered to MR images using a nonlinear warping algorithm (other components). A pathologist who was blinded to the classifier results manually segmented digitized histological images. The sensitivity of the classifier, as determined by pixel-by-pixel comparison with the pathologist's segmentation and micro-CT, was 60.4% for fibrous tissue, 83.9% for necrosis, 97.6% for calcification, and 65.2% for loose connective tissue. The corresponding values for specificity were 87.9%, 75.0%, 98.3%, and 94.9%, respectively. In conclusion, multicontrast MRI was successfully used in conjunction with a supervised classification algorithm to identify plaque components in endarterectomy specimens. Furthermore, this methodology will provide a framework for comparing different classification algorithms, and determining which combination of MR contrasts will be most valuable for in vivo plaque imaging.

Arteriosclerosis↗
PubMed2001

The use of classification and regression trees in clinical epidemiology.

A critique is presented of the use of tree-based partitioning algorithms to formulate classification rules and identify subgroups from clinical and epidemiological data. It is argued that the methods have a number of limitations, despite their popularity and apparent closeness to clinical reasoning processes. The issue of redundancy in tree-derived decision rules is discussed. Simple rules may be unlikely to be "discovered" by tree growing. Subgroups identified by trees are often hard to interpret or believe and net effects are not assessed. These problems arise fundamentally because trees are hierarchical. Newer refinements of tree technology seem unlikely to be useful, wedded as they are to hierarchical structures.

Algorithms↗
PubMed1998

Classification of otitis media.

OBJECTIVES: To review the International Classification of Diseases (ICD) system of classifying otitis media. To highlight the failings of this system and to propose an alternative (Read version 3). STUDY DESIGN: An historical review of the literature and a presentation of two distinct classification schemes. METHODS: The advantages and disadvantages of the alternative schemes are analyzed and an algorithm for the classification of otitis media is presented that describes the decisions taken to reach a diagnosis. RESULTS: The proposed classification of otitis media as used in the Read Codes thesaurus is more logical than the ICD classification. CONCLUSIONS: The widely used ICD schema of otitis media fails to provide the otologist or general medical practitioner with a logically organized set of terms to describe inflammatory middle ear disease. This article proposes a simple, hierarchical classification that can be used by specialists, generalists and epidemiologists alike.

Acute Disease↗
PubMed2006

Phyloproteomics: what phylogenetic analysis reveals about serum proteomics.

Phyloproteomics is a novel analytical tool that solves the issue of comparability between proteomic analyses, utilizes a total spectrum-parsing algorithm, and produces biologically meaningful classification of specimens. Phyloproteomics employs two algorithms: a new parsing algorithm (UNIPAL) and a phylogenetic algorithm (MIX). By outgroup comparison, the parsing algorithm identifies novel or vanished MS peaks and peaks signifying up or down regulated proteins and scores them as derived or ancestral. The phylogenetic algorithm uses the latter scores to produce a biologically meaningful classification of the specimens.

Algorithms↗
PubMed2002

Bounded-depth threshold circuits for computer-assisted CT image classification.

We present a stochastic algorithm that computes threshold circuits designed to discriminate between two classes of computed tomography (CT) images. The algorithm employs a partition of training examples into several classes according to the average grey scale value of images. For each class, a sub-circuit is computed, where the first layer of the sub-circuit is calculated by a new combination of the Perceptron algorithm with a special type of simulated annealing. The algorithm is evaluated for the case of liver tissue classification. A depth-five threshold circuit (with pre-processing: depth-seven) is calculated from 400 positive (abnormal findings) and 400 negative (normal liver tissue) examples. The examples are of size n=14,161 (119 x 119) with an 8 bit grey scale. On test sets of 100 positive and 100 negative examples (all different from the learning set) we obtain a correct classification close to 99%. The total sequential run-time to compute a depth-five circuit is about 75h up to 230h on a SUN Ultra 5/360 workstation, depending on the width of the threshold circuit at depth-three. In our computational experiments, the depth-five circuits were calculated from three simultaneous runs for depth-four circuits. The classification of a single image is performed within a few seconds.

Algorithms↗